Science can be a powerful tool for discovery and problem solving, but it can also be a messy, nonlinear process that does not provide all the answers. In such cases, people are forced to make the best decisions possible based on the available information, often erring on the side of caution—especially where public health is concerned. Whether or not those decisions are too conservative will often be the topic of long debate. One such example, involving drinking water standards and the health of newborns, has continued for more than half a century.
In 1987, the Journal of the American Medical Association reprinted, as part of its Landmark series, a case study originally published in 1945, accompanied by a commentary and a report of a new and fatal instance. The author of the original paper, Hunter Comly, was a pediatric resident in Iowa City when he described two examples of a previously unrecognized blood condition in infants. Called infantile methemoglobinemia, the affliction had as its main symptom cyanosis, or turning blue (thus the condition was also sometimes called blue baby syndrome). Cyanosis can also be a symptom of a congenital heart disease, so Comly felt that the two conditions might be confused.
In the congenital condition, called tetralogy of Fallot, a complicated structural defect allows blood returning from the body to the heart to be pumped out again without going to the lungs to be resupplied with oxygen. The hemoglobin in red blood cells turns red when iron in the molecule binds to oxygen; it turns bluish-purple when the oxygen is unloaded, which is why veins are bluish. With too much deoxygenated blood in the arteries, the skin turns from pink to blue (cyanosis comes from the Greek kyanos, meaning dark blue).
Alternatively, some chemicals can oxidize the iron in hemoglobin. The altered form, called methemoglobin, loses the ability to bind oxygen, and the pigment now changes to greenish brown or almost black. The human body contains enzymes to reverse methemoglobinemia, but only up to certain levels. After blood levels reach 15 percent, adults become visibly cyanotic. If more than half of the hemoglobin is converted, oxygen transport, particularly to the brain, is severely hampered, respiratory distress is likely, and death is possible.
In congenital heart disease the cyanosis is apt to improve if treated with oxygen, and if a sample of shed blood is shaken in air, it often becomes lighter and redder in color. Since methemoglobin does not bind oxygen, infants with that condition do not “pink up” on oxygen, and samples of shed blood exposed to air undergo little or no change in color.
The two conditions also look a little different. During the influenza epidemic of 1918, the word “heliotrope” was used to describe the color of the cyanosis of hypoxia, whereas the cyanosis of methemoglobinemia is often described as slate-gray. Simple laboratory tests are available to identify and quantify methemoglobin in blood, and infants with high levels of it respond rapidly to intravenous methylene blue—this blue dye turns the blue baby pink. When cyanosis develops post-partum, it is usually noticed first in the lips, spreading gradually to the nail beds of the fingers and toes, the face and then the whole body. Both congenital heart disease and methemoglobinemia tend to result in hypoxia and peripheral vasodilation, which may intensify the cyanosis.
Don’t Drink the Water
In the condition first described by Comly (now known as well-water methemoglobinemia), affected infants appeared healthy at birth and remained so through discharge. The cyanosis arose only after days, weeks or months in the home environment. What made Comly’s paper so noteworthy was the full explanation offered for the etiology of the disease—one that has been largely undisputed for 60 years.
He recognized that “the only significant change in the infant’s environment from hospital to farm home was in the water that was used to prepare the formula.” But what was the nature of the difference between hospital and home water? After the third episode of admission and treatment with methylene blue for one child, her father demanded to know exactly what “poison” in his well was responsible. Analyses of the water showed high levels of nitrate.
Inquiries to other physicians in the area about unpublished, similar cases turned up 17 more, with one fatality, suggesting that the condition was more common than suspected. As the number of physicians who reported seeing unexplained cyanosis in infants who had been fed powdered formula prepared with well water grew, measurements of the nitrate content of Iowa wells showed that more than half of the 91 wells initially sampled contained nitrate nitrogen in excess of 10 parts per million, and 20 contained more than 65 parts per million.
Those wells that tested high for nitrates were clearly undesirable from a public health standpoint. Most of them were old, shallow and dug rather than drilled, and they often had inadequate casings. Many were poorly covered so that surface runoff containing nitrate fertilizers or animal excreta rich in nitrates could enter freely. They were often found near barnyards, feedlots or pit privies, and some were contaminated with coliform bacteria. Although the wells served as the drinking water source for the entire family, only formula-fed infants were affected, and most were less than three months of age.
Subsequent independent surveys in midwestern states confirmed the initial results in Iowa. The most complete survey of all 48 states, plus what were then the territories of Alaska and Hawaii, was compiled in 1951 and identified 278 cases with 39 deaths. No cases were found in which wells contained 10 parts per million nitrate nitrogen or less, and only 2.3 percent of the cases involved wells with between 10 and 20 parts per million. Above 20 parts per million, the severity of the symptoms seemed to parallel the amount of nitrate present. Breastfed infants were never involved; neither were families who used municipal water supplies. No reports of cases prior to World War II were found. However, physician reporting was not mandatory for the condition, and the true number may have been much higher.
But by the time that report was published, the worst seemed to have passed, and the number of cases fell off steadily through the early 1950s. Today the disease has all but disappeared, with reports appearing only sporadically in the literature. Only two cases have been reported since the mid-1960s and none since 2000. Within 10 years the epidemic had waned as suddenly as it had appeared, without any preventive action having knowingly been taken. Whether it was because of public awareness, a massive improvement in rural drinking water quality, a trend toward breastfeeding or other factors may never be known.
Nitrates to Nitrites
Comly realized that he needed to explain the sensitivity of newborns to nitrates in order to complete the picture. He also needed an explanation for how biologically inert nitrate was responsible for oxidizing hemoglobin to methemoglobin. The most plausible candidate for the oxidizing agent was nitrite, which was known to be a potent generator of methemoglobin. To find it, he needed to look no further than the diverse normal microflora of the infant gut. Many common organisms found in the bowel were known to convert nitrate to nitrite; the nitrite could then be readily absorbed into the blood to do its damage.
In adults, however, ingested nitrates—more than 97 percent of which come from vegetables and other foods, not water—never come in contact with intestinal microflora, which are found largely in the colon. Nitrates are absorbed rapidly in the duodenum and excreted promptly in the urine. That led to a consideration of the second-most common finding in affected infants, namely evidence of gastrointestinal upset in the form of vomiting or diarrhea, things so usual as to excite little attention. A few infants who were tested were found to have high gastric pH in relation to adult values. Presumably the acidic milieu of the stomach and duodenum discourages microflora from invading the small intestine of adults. But if the pH were high enough in newborns, perhaps nitrate-reducing organisms could survive in the small intestine and come in contact with unabsorbed nitrate. This problem should be self-solving as infants mature, because acid production in their stomachs increases to adult levels by about six months of age. That coupled with a relative deficiency in the enzyme that reverses methemoglobin production in infant red blood cells could be enough to explain their sensitivity to nitrates. As Comly’s explanation stands, it is both complete and plausible, but a long way from scientifically proven.
Largely because of Comly’s recommendation, 10 parts per million of nitrate nitrogen quickly became the standard for potable water supplies in the U.S., and that maximum was reaffirmed in 1995 in the most recent review by the National Research Council. Unfortunately, nitrate removal from raw drinking water involves considerable expense. Conventional treatment processes are ineffective. A much more expensive process of ion exchange is the only one currently used. But those at risk—infants under three months of age—are one of the most vulnerable and emotionally charged groups in our population, one that all agree we would most like to protect.
A New Syndrome?
In the 1980s a new phenomenon seemed to appear, namely, a form of infantile methemoglobinemia associated with inflammatory bowel disease (including diarrhea, acidosis, infection and gastroenteritis), where exposure to excessive nitrates could not be clearly established. The first of these reports appeared around 1982. Over the course of a year the authors had seen 11 infants with the triad of methemoglobinemia, diarrhea and acidosis. All were under three months of age, and all lived in homes with municipal water supplies. Although infection was suspected in at least some of these cases, no common fecal pathogens were isolated.
In the same year, a report from Israel described a study of 58 infants, age one week to one and a half years, admitted to the hospital for acute diarrhea. Over 100 admitted infants without gastrointestinal disturbances served as controls. All were fed the same milk, food and water, which had been analyzed for nitrates and nitrites. The study excluded infants with confounding factors such as genetic defects or exposure to drugs or chemicals that generate methemoglobin. None of the affected infants was said to be cyanotic, and only 12 of the 58 had methemoglobin levels above 8 percent. However, infants with the more severe diarrhea seemed to have higher methemoglobin levels. The nitrate concentration of the drinking water was well under the U.S. standard, and the diet was unusually low in nitrates. There was no clear correlation of the degree of acidosis and methemoglobin levels.
The most seminal observation was that the affected infants had high blood levels of nitrate, and those levels did appear to parallel the degree of methemoglobinemia. Affected infants excreted several times more nitrate in their urine than they consumed in their diet. In the control group the urinary excretion of nitrate was about the same as the daily intake or only slightly higher. Previously, scientists at the Massachusetts Institute of Technology had demonstrated nitrate production within the body in germ-free rats.
As is often the case, these two reports opened the floodgates. Between 1983 and 1996 reports of more than 90 cases came out in the medical literature, and then, maddeningly, the reports stopped appearing, just as they had with well-water methemoglobinemia 50 years before.
In 1999, Alex Avery of the Center for Global Food Issues at the Hudson Institute and his colleagues offered an ingenious explanation of the etiology after an analysis of some seemingly unrelated literature on inflammatory bowel disease. Noting that the earlier Israeli study strongly suggested that nitrites produced in the body were responsible for the methemoglobinemia and that their end product, nitrates, were elevated in urine and blood, they suggested the involvement of endogenous production of nitric oxide from arginine. The importance of nitric oxide as a normal biological mediator in a number of physiological processes has only recently been recognized. In solution, nitric oxide exists in equilibrium with nitrite. There is some published evidence to suggest that certain viral or bacterial infections of the bowel cause an increase in nitric oxide levels in human colonic epithelial cells. Although these studies were focused on explaining the pathologic changes in the bowel, they offer a possible explanation for the methemoglobinemia. That explanation, however, does not yet suggest why infants may be more susceptible.
At a Standstill
Few pursuits are more frustrating to the biomedical scientist than attempts to investigate diseases of very low incidence. It is a mixed blessing that both episodes of infantile methemoglobinemia ended spontaneously, because at this juncture no further elucidation of either is in sight. Try as we might, we face formidable obstacles in studying diseases that no longer exist and for which there are no satisfactory animal models. Nor is there any incentive to carry out such studies.
The hypothesis that Avery has offered for the etiology of the endogenous disease is, of course, incompatible with the exogenous well-water nitrate hypothesis. If Comly was wrong, exogenous nitrates were a red herring, and the drinking water standard for nitrate has been set at an unnecessarily low level. Communities may have incurred needless expense in pursuing nitrate removal. However, the evidence for the involvement of well-water nitrate in the first miniepidemic is at least as strong, if not stronger, than the evidence for the involvement of gastrointestinal infections in the second. The clear dose-response relationship between the well-water nitrate content and the severity of the methemoglobinemia, and the logical explanation for the sensitivity of infants to nitrate, support the etiology proposed for the first episode, but their equivalents are lacking for the second.
Science is simply not always able to provide neat and clean answers, and in order to protect the public, expensive policy decisions must sometimes be made based on whatever facts are known. We seem to be forced to the conclusion that an exceedingly rare toxic condition, methemoglobinemia in infants, is linked to two episodes of exposure to endogenous nitrite, but generated by two entirely different mechanisms. More improbable still is that both episodes suddenly appeared and then spontaneously resolved over the space of a dozen years, each in the second half of the 20th century. As Sherlock Holmes famously remarked to Dr. Watson, “When you have eliminated the impossible, whatever remains, however improbable, must be the truth.” It also may be, in a way, a vindication of the now 50-year-old drinking water standard for nitrate.